A unified machine learning approach to time series forecasting applied to demand at emergency departments

Vollmer, Michaela A.C.; Glampson, Ben; Mellan, Thomas; Mishra, Swapnil; Mercuri, Luca; Costello, Ceire; Klaber, Robert; Cooke, Graham; Flaxman, Seth; Bhatt, Samir

BACKGROUND: There were 25.6 million attendances at Emergency Departments (EDs) in England in 2019 corresponding to an increase of 12 million attendances over the past ten years. The steadily rising demand at EDs creates a constant challenge to provide adequate quality of care while maintaining standards and productivity. Managing hospital demand effectively requires an adequate knowledge of the future rate of admission. We develop a novel predictive framework to understand the temporal dynamics of hospital demand. METHODS: We compare and combine state-of-the-art forecasting methods to predict hospital demand 1, 3 or 7 days into the future. In particular, our analysis compares machine learning algorithms to more traditional linear models as measured in a mean absolute error (MAE) and we consider two different hyperparameter tuning methods, enabling a faster deployment of our models without compromising performance. We believe our framework can readily be used to forecast a wide range of policy relevant indicators. RESULTS: We find that linear models often outperform machine learning methods and that the quality of our predictions for any of the forecasting horizons of 1, 3 or 7 days are comparable as measured in MAE. Our approach is able to predict attendances at these emergency departments one day in advance up to a mean absolute error of ±14 and ±10 patients corresponding to a mean absolute percentage error of 6.8% and 8.6% respectively. CONCLUSIONS: Simple linear methods like generalized linear models are often better or at least as good as ensemble learning methods like the gradient boosting or random forest algorithm. However, though sophisticated machine learning methods are not necessarily better than linear models, they improve the diversity of model predictions so that stacked predictions can be more robust than any single model including the best performing one.